Internat. J. Math. & Math. Sci. VOL. 12 NO. (1989) 123-130 123 ON INTEGERS n WITH Jt(n) < Jt(m) For m > n J. CHIDAMBARASWAMY and P.V. KRISHNAIAH Department of Mathematic-s University of Toledo Toledo, OH 43606 (Received April I0, 1987) ABSTRACT. Let F all m > n, Jt(n) as F t be the set of all positive integers n such that < Jt(m) for In this paper, it being the Jordan totient function of order t. been proved that (I) every postive integer d divides infinitely many members of n as n in F (3) every (2) if n and n’ are consecutive members of F t, t prime p divides n for all sufficiently large n Ft(x) is the number of n KEYS WORDS AND PHRASES. g F t F g t and (4) Log Ft(x) t << log x where x. that n Euler totient function, Jordan totient function of order t. IIA,N. 1980 AMS SUBJECT CLASSIFICATION CODE. I. Jr(n) INTRODUCTION. In [I] Masser and Shlu consider the set F of positive integers n such that (n) < (m) for all m (n) being the Euler totlent function. > n, Calling members of F sparsely totient numbers, they prove, among other results, that (I) every integer divides some member of F (2) every prime divides all sufficiently large members of F (3) the ratio of consecutive members of F approaches (4) log F(x) << log1/2x where and F(x) is the counting function of F; that is, the number of members of F which do not exceed x. In this paper, using similar methods, we extend the above results to the set F t Jt(n) of all positive integers n such that < Jt(m) well known Jordan totlent function of order t [4]. as the number of incongruent t-vectors (a it being understood that t-vectors (a I, if a i bi(mod g i g t and that n) for Jt(n) Moreover this function n t Jt(n) (I p a at) Jt(n) t for all m > n, Jr(n) being the Jr(n) is defined that ((a at)n) We recall that mod n such and (b b t) I, are congruent mod n is given by the formula -t) coincides with Cohen’s [3] generalization Euler totient function, defines as the number of positive integers a n (I.I) @t(n) t of the such that J. CHIDAMBARASWAMY AND P.V. KRISHNAIAH 124 t (a,n) where (x,y) t Clearly denotes the largest t th power common divisor of x and y. t l(n)) (and hence Jl(n) is the same as (n). Denoting by (n) the number of distinct prime factors of n, we order these prime P1 > P2 factors as > P thus (n)" Pr Pr(n) is the r th largest prime factor of Likewise we order the primes not dividing n as n. Pr > for the r th prime in the ascending n, we write n, < Q1 Q2 < Further we write sequence of all primes. For positive integral to denote the quotient of n by its largest square free divisor. positive integer u, we write a t x t+u + (t+u) x For a for the unique positive root of the equation t u (1.2) 0. It may be directly verified that 2u t+u Ft(x) Finally we write 2. =< a u < (1.3) for the number of members of F t that do not exceed x. MAIN RESULTS. We prove the following results: Let k THEOREM 2.1. d < I, 2, d 0 be integers such that (2.1) Pk+l and d Then d t t .1. tF Every prime p divides n for all sufficiently large n tLet u be a fixed positive integer. As n in F we have t THEOREM 2.2. THEOREM 2.3. log-In (a) lim inf Pu (n) (b) lim sup Pl(n) log-ln (c) au (d) (2.2) is a member of F tEvery positive integer d divides infinitely many members of F t. -I If n and n’ are consecutive members of F t, then n n as COROLLARY 2.2. in F I). PI’’" Pk-lPk+ COROLLARY n (d+l)t(p 1) < (Pk+ llm inf n llm sup n 2 Qu(n) log-In llm sup n (n) log-In (2.3) au-I Pt+u(n) log-ln (2.4) and (e) lim sup n Pl(n) log-t-In THEOREM 2.4. 3. log (2.5) Ft(x) << logx. FOR THE PROOFS OF THE THEOREMS WE NEED THE FOLLOWING LEMMAS. LEMMA 3.1. Let r be a positive integer and numbers satisfying (i) (ii) S t Xl’’’Xr x < yl...yr x y" i Yi Then and y (Xl-l) x i Xl,...,Xr, for i (x r-l)(x-l) < YI’’’’’Yr’ x and y be real r and (Yl-l)...(Yr-l)(y-l). ON INTEGERS n WITH Jr(n) Jr(m) FOR m > n 125 This is lemma 3.1 of [6]. For positive integers a and b we write f(a,b) to denote the smallest multiple of b Clearly that exceeds a. a < f(a,b) _-< a LEMMA 3.2. If n e F then k-2t < re(n) PROOF. n < (3.2) Pk Pk+l Pl k. The second inequality in (3.2) implies that m(n) possible, that 0 < m(n) f(n, (I p i-t) < m <_-I n On the other hand, since Pl Suppose if k. -t (I (3.3) Pk_2t Pk-t )’ we have, by + + Pl" n "Pk-t Pl -< It follows from (I.I) that k- 2t. Jt(n)m-t Choosing m (3.1) b. and t Pk Pl + (3.1) and (3.2), + < Pk-t+l" "Pk (3.4) Pk-t divide m, we have, in virtue of (3.3), Pk-t -t -t -t Jt(m)m -<- (l-Pl (l-pk-t) -<- Jt(n)n-t (l-Pk-t2t+l) =< t Jt (n)n-t (l-pk-tt) (l-Pk-t) so that, (3.4) now yields Jt (m) Jt(n) + < (I -t Pk_t) t -t (I Pk_t) < Hence the lemma follows. contrary to the hypothesis that n e F t REMARK I. Since for each n there is a unique k such that (3.2) holds, lemma 3.2 implies that m(n) in F as n t eu Pt+u< t + u then LEMMA 3.3. If u is a positive integer, n e F t and m(n) -I f(a,b) so that and na m PROOF. We write a b QI Qu PI + n Since Qu Q1 PI Pt+u so that Pt+u’ b + u Qu p-t-u t+u (3.5) are prime factors of m but not of n where as the prime factors of n may or may not divide m, we have Jt(m)m -i (l-Qut)( l-p-t1 (l-Qlt) (3.5) and the hypothesis that n < Jt(m)(Jt(n))-I Qu" < F t uP-t-u.t+u u (l+QU (l+Qu p-t-u, t+u u t I--t -I t(n)n t+n) J (l-p?t) -I together imply i=l < -I t- (l-Qi (I- -t u Qu t+u i=l (I -t -Pt+u -t-u -t 126 J. CHIDAMBARASWAMY AND P.V. KRISNAIAB Taking (t + u) th roots and employing the well known inequality x r <- + r(x-l) for x > 0, 0 < r < I, we obtain I-Q: < l-uQ t (t+u)-I < l-uQ Cancelling 1+Q.U < -t -I (t+u) u + u -t-u Pt+u (tdm)-I tQu (t+u)Q in the above and multiplying by t(Qu t+u -I Pt+u + (t+u) (Qu When t -I t Qu Pt+u Pt+u which, in virtue of (1.2), implies that REMARK 2. Pt+u-t-u (t+u)-I) u l+tQu u > 0 -I this lemma 4 of u we arrive at > u [I] and when this yields a t slight improvement of lemma 7 of [I]. m t(Ql(n))t+l. LEMMA 3.4. For n e F t we have Pl(n) < PROOF. We write P for Pl(n) and Q for -I we see n f(P,Q)P that Ql(n)" Suppose P -> t Q t+-+ choosing +_=_<o +__l n (3.6) tQt Qt+l Arguing as in lemma 3,3, we have, since P Jt (m) m-t < by our assumption, l-p-t)-I Jt (n)n-t (1-Q-t) (l_Q-t)(l_Q-t(t+l)) + Q_t (l+Q-t + Jt(n)n -I 2 -t Jt(n)n -t so that by (3.6) Jt (m) (Jt (n) )-I contrary to the hypothesis. (l+Q-t + + Q-t 2)-I (l+t-IQ-t)t .< I, This establishes the lemma. For n F t, n, Writing m n f (n,,Q)n, LEMMA 3.5. PROOF. .< < t(Ql(n)) t+l we note that +Q <m<= n n, so that, since n e exp(O) Ft, Jt(m)(Jt(n))-I <= t (l-Q-t)(l+Qn, I) < exp (-Q-t)exp(tQn, I). The lemma follows on comparing the exponents. LEMMA 3.6. A < p Let A > 0, M > 3 and (A;M) be the number of primes p with A+M ((A;M) (A+M) w(A)). Jr(n) ON INTEGERS n WITH < Jr(m) 127 FOR m> n Then <- (A;M) 2M(log M) -I {i + 0 (log log M log-IM)} and the 0-constant is independent of A. This is Theorem 4.5, Chapter 19 of [2]. REMARK 3. Lemma 3.6 implies that (A;M) 3M log M for all A 0 and sufficiently large M. 4. PROOFS OF THEOREMS. PROOF OF THEOREM 2.1. (2.2). Let n d pl...Pk_l From (I.I) and (2.2) we have t Jt (n) <= dt (Pl -I) t (d+t)t(pl-I) < Let m > n. t (t Pk_l)(Pk+ Jt(m) -> m t t s-> Case (I) k+l. (plt_l > (Pl-l) "> Jt(n) <= s (l-ps (Ps & m < Pl Ps+l and the last -t t-I (4.2) t (pkt-l)(Pk+l-l) t t (Pk-l)(d+l) t by (2.1) by (4.1) k, (m) .< k-l. Jt(m)m-t Ps We have, by (4.2), Jr(m) -> Case (2) -t (Pl t_l >- Pl (4 I) Hence s. (l-Pl -I) (Pk-l- I) (pkt-I There is a unique s such that inequality implies that m(m) where d,k, satisfy (2.1) and Pk+ In this case (l-Pl -t) -t (I Pk-I where as -t Jt (n)n-S (l-p t) (1-plt) so that Jt(m) Case (3) > -t (l-Pk_ I) (l-Pk+) --t (l-Pk_ I) Jr(n)" s m(m) and k t Jt(m) t t (Pl-l) __> (d+l) t > d+l. In this case s (pt-l) m, m, m, => Jt (n) t (Pl-l) t (Pk-l) t (Pk-l) k (m) and 128 J. CHIDAMBARASWAMY AND P.V. KRISHNAIAH in virtue of (4.1). Case (4) -< s (m) and k m, =< q k-l, Y l Taking r > qk-I qk ql t Pl Pk-I q y x t t p i (qk_l)(qk_l) (ql_l) > ql "> Then m, (d Pk+g and qk’ ql ’s m, ql Let m d. distinct prime factors of m in ascending order. x d -I t t -t Pk+gm* (P-I (p-l) Pl being the and since m > n we have I) (d in lemma 3 we obtain -t Pk+ m,-I) t t so that (Pk_l-l) (d > dt "> by (4.1). t (Pk_l-l)(Pk+ -I) Jt (n) This completes the proof of Theorem 2.1. PROOF OF COROLLARY 2.1. Let d be any positive integer. that (2.1) holds, we can take each k t t (Pl-l) pk+-m, 2 such that 0 so that (2.2) holds. For each k Thus d 2 such Pk Pl Ft for > d+l. Pk+l As the proof of Corollary 2.2 is essentially the same as that of the Corollary given in section 3 of [I], we omit it. PROOF OF THEOREM 2.2. Pr > pal -I Let p be a given prime. Choosing r such that we see that (n) E r + t + for all n in F such that n Pl t Pr+3t For such n we have in virtue of lemma 2. > P Pt+l Pr al -I so that > QI > p Ul Pt+l in virtue of lemma 3.3 (u REMARK I. I), yielding that pln. Though this theorem follows immediately from Theorem 2.3 a direct proof seems desirable. (a) For any n e PROOF OF THEOREM 2.3. Pl n < Pk (n) >- k By lemma 3.2, Pu(n) P-u+l (e(x) x), Pk ffi> Pk-2t-u+2 log (Pl Pl 2t Ft, there is a unique integer k satisfying (4.3) Pk+l + so that, for a fixed integer u, in F log n as n t slnce, by the prime numbers theorem log n Pk and Pk-2t-u+2 Pk as k (hence as n in Ft). ON INTEGERS n WIT p (n) u >- Thus lim inf n log n P 0 in theorem 2.1), we have P (n) P u (n) log n --Pk" Hence through such members of F t P (n) u I. lim inf log n as n -> (b) For k >- 2, t Pk+ I) says that 0 theorem 2.1 (with d t < 2 129 FOR m > n On the other hand, considering members n of F t of the form 1. l, (take d Pk Jr(n) < Jr(m) (p-l) + => Ft" Pk-I Pk+ Pl to be the largest subject to the above condition we have choosing t Pk+ +I > 2 t t + (Pk-l) t > (4.4) Pk+ so that Pk++l where n- Now Pl Pl(n) < 2(Pk-1) 2 log n Pk-lPk+ since n satisfies Pk+ Pk++l and this yields p1(n) 2 lim sup lira inf n >- Io= n (c) Since, by lemma 3.3, Qu Qu (n) (4.3) in virtue of (4.4). > au Pt+u -> a log n (a) we have, from u On the other hand, choosing k as in (4.3) we see that QI <= for all n in F t, where as, for members of F t of the form P Since log n we have Pk+u Pk+l Qu < Pk+u have Qu Pk+u* and hence Pk we Pk lira sup n Qu (n) log n i. (d) and (e) now follow by applying lemmas 3 and 4 respectively. number of members n of PROOF. OF THEOREM 2.4. We write G(x) Ft(x) x < the theorem follows from which F satisfying n -< x and show that log G(x) Ft() log1/2x t easily. Throughout the proof we assume that x is a sufficiently large positive real number. We write u [logx t and note that t+u _-> 2 Putting Q1 P+I we conclude that + (log log x) 2t. -I] For n e and noting that + (4.5) t Ft, x n > -> _34 we have ZQl(n) log Q1 (n) Ql(n) 4 i log n by (2.3) by the prime number theorem 130 J. CHIDAMBARASWAMY AND P.V. KRISHNAIAH flog log}x n > log log n >t+u log log x Each n is specified uniquely when n, 2 t +2t. and the prime factors of n are given. We estimate an upperbound for G(x) by estimating upper bounds for the number of choices, x F consistent with n x], for each of (a) n, (b) the largest t+u prime factors t of n, namely and (c) the prime factors of n that are less that (, au Pt+u Pl’’’’’Pt+u Pt+u’ Pt+u (d) the prime factors of n that lie in [a u )" (a) By lemma 3.5 and (2.3) the number of cholcesfor n, does not exceed t+l t+l t (2 log x) and hence does not exceed (2 t log x) t+l -< i <- t+u in virtue of (1.8) so that the "< <" 2t (log x) for (b) Pi PI number of choices for Pt+u does not P1 (c) Each choice of QI’ factors of n that are < a u to at least one choice for Qu P exceed (2t log x)(t+l)(t+u). gives rise to exactly one choice of the prime and each choice of these prime factors gives rise t+u au Pt+u < Qu and all primes in (I, au Pt+u {QI Qu divide n. If P1 Pr then r Pr P" "< 2t logt+Ix, "< "< and (3t lolgt+Ix) z so that the Qu Pr+u" Hence Qi <- Pr+u < (r+) Q1 Pr+l QI’""Qu since < 2 number of choices for does not exceed (3t (d) Let M M ’Qu Q1 and hence for the prime factors of n in logt+Ix) 2u. Pt+u (I u Pt+u and (l,auPt+u) note that u Pt+u <" 2t(t+u) -I Pt+u -< 3t(t) -I a -I log x u < 4t a -1 U log1/2x log log x in virtue of (1.3), (2.4) and (4.5) respectively. lemma 3.6, the number of primes in [au Pt+u’Pt+u Hence, by remark 3 following does not exceed 24t au-I log1/2x and consequently the number of choices for the prime factors of n that lle in this interval does not exceed 2 24tu-llog x. 3 (t+u) (t+l) (3t log x) Combining the estimates (a) through (d) we obtain G(x) -1 exp (24 t aU log 2) from which the desired order estimate follows in virtue logx of (4.5). It would be interesting to investigate whether results of this nature are available for more general totients; e.g. totients with respect to a polynomial [3] and with respect to a set of polynomials [4], introduced by the first author. REFERENCES I. MASSER, D.W. and SHIU, P. (1986), 407-426. 2. HUA, L.K. On sparsely totient numbers, Pacific J. of Math. 121 Introduction to Number Theory, Springer-Verlag, Berlin-Heidelberg-New York, 1982. 3. CHIDAMBARASAMY, J. Totients with respect to a polynomial, Indian J. of Pure & Applied Math. _5 (1974), 601-608. 4. Totients and unitary totlents with respect to a set of polynomials, Ibid, I0 (1979), 287-302.